Professional Documents
Culture Documents
Visit analogdialogue.com
In This Issue
Share on
Introduction
In systems with multiple supply voltages, operational amplifier power Table 1 shows the results of all AD8616 pins in such conditions. Before V+
supplies must be established simultaneously with or before any input is applied, the voltage at the V+ pin and OUT pins is negative. This may
signals are applied. If this doesnt happen, overvoltage and latch-up not damage the op amp, but if these signals are connected to terminals
conditions can occur. on other chips that havent been fully powered (for example, assuming
the ADC uses the same V+, and its power pin normally tolerates only
However, this can sometimes be a difficult requirement to meet in real- 0.3 V minimum voltage), the chips may suffer damage. A similar issue
world applications. This article takes a look at the activity of op amps in will happen if V+ is powered up before V.
different power sequence situations (see Table 2), analyzes possible
issues, and presents some suggestions. Table 2 highlights some possible situations in power sequencing.
positive supply (negative power is present and positive power is absent). Figure 1. AD8616 test circuit with 3 V V applied and V+ absent.
Pin1: OUTA Pin2: INA Pin3: +INA Pin4: V Pin5: +INB Pin6: INB Pin7: OUTB Pin8: V+
0V Present Absent No No
Case 2
0V Absent Present No No
V+
Evaluation Setup
Figure 3 is used to measure the activity of the op amp. Channel A and
D3 V+
IN Channel B are each configured as a buffer, and the Channel B noninverting
V+ D4 D7 input is connected to the GND by a 100 k resistor. By making V+ absent
D5 D6 OUT
D1 D8
(V present) or V+ present (V absent), the input and power-related vari-
V
+IN ables can be measured by the ampere and voltage meters. By analyzing
D2 V these variables, we can determine the current flow path.
ADA4077
V
V ISY+ A V
V+ IOUT 100 k
A V+
V+
V+
IN A V+ IN B
D3 V+
OUT A OUT B
IN OVP IB+ A B
V+ +IN A 100 k +IN B
D4 D7
A
D5 D6 OUT V V 3 V
D1 D8 V
+IN OVP
D2 V
ISY A V
ADA4177
V
V
Figure 3. Setup for power sequencing test.
Figure 2. ADA4077/ADA4177 simplified block diagram.
Case 1: Input Is Floating
Table 3. Internal Diode of an Op Amp Table 4 shows the results of a floating input and one absent supply. When V
is present and V+ is absent, there is a negative voltage at the V+ pin. When
ADA4077 ADA4177 V+ is present and V is absent, there is a positive voltage at the V pin.
D1 0.838 Unknown
D2 0.845 Unknown Testing the ADA4077-2 and ADA4177-2 reveals similar results. No large
D3 0.837 Unknown
currents are observed at the input pins and power pins, and the op amp
with floating input remains safe when a power rail is absent.
D4 0.844 Unknown
D5 Unknown Unknown
Case 2: Input Is Grounded
D6 Unknown Unknown
Table 5 shows the results when the input is grounded. Note for IB+, a
D7 0.841 0.849
negative value means the current flow out of the +IN terminal. For IOUT,
D8 0.842 0.849 a negative value means the current flow out of the IN terminal.
V+ V+
1.6 mA 0.2 mA
V+ V+
D3 V+ D3 V+
IN IN OVP
0.846 V V+ D7 V+ D7
D4 11.99 V D4
0.7 mA OUT D5 D6 9.1 mA OUT
D1 15 V D5 D6 D8 2.68 V 9.1 mA D1 D8 11.98 V
0.7 mA
+IN +IN OVP
D2 15 V D2 V
2.3 mA 9.3 mA
0V ADA4077 2.3 mA 0V ADA4177 9.3 mA
15 V V
15 V 15 V
V+ V+
1.65 mA 0.11 mA
V+ V+
D3 V+ D3 V+
IN IN OVP
9.14 V V+ D7 V+ D7
D4 9.09 V D4
D5 D6 0.75 mA OUT D5 D6 8.75 mA OUT
9.99 V D1 15 V D8 7.3 V 9.92 V 8.75 mA D1 D8 9.06 V
0.75 mA
+IN +IN OVP
D2 15 V D2 V
2.4 mA 8.86 mA
ADA4077 2.4 mA ADA4177 8.86 mA
15 V V
15 V 15 V
Figure 5. ADA4077/ADA4177 current path when V+ is absent (10 V input).
Table 6
Condition V+ V ISY+ (mA) ISY (mA) IB+ (mA) IOUT (mA) IN (V) OUT (V)
ADA4077-2 All power 15 15 1.03 1.01 0.00098 0.00003 10 9.97
V+ absent and positive input 9.14 15 0 2.4 2.396 1.653 9.99 7.3
V+ absent and negative input 10.83 15 0 2.41 2.308 1.651 10.02 12.66
V absent and positive input 15 10.83 1.81 0 1.689 1.055 10.02 12.09
V absent and negative input 15 9.15 1.77 0 1.759 1.031 9.99 7.88
ADA4177-2 All power 15 15 1.02 1 0.00099 0.00009 9.99 9.97
V+ absent and positive input 9.09 15 0 8.86 8.866 0.113 9.92 9.06
V+ absent and negative input 12.33 15 0 4.31 4.18 0.039 10.02 12.32
V absent and positive input 15 11.42 1.33 0 1.2 0.056 9.99 11.43
V absent and negative input 15 8.33 1.51 0 1.492 0.062 9.97 8.32
V+ V+
6.2 mA 2.1 mA
V+ V+
D3 V+ D3 V+
IN IN
8.77 V V+ D7 2.389 V V+ D7
D4 D4
D5 D6 OUT D5 D6 OUT
D1 15 V D8 23 mA 1.373 V 9.98 V +IN D1 15 V D8 3.9 mA 0.284 V
+IN 24 mA 4.7 mA
29.2 mA
15 V 47 1 k 15 V 47
6 mA
30.2 mA D2 6.8 mA D2
9.98 V ADA4077 1 mA ADA4077 0.8 mA
15 V 15 V
15 V 15 V
Figure 6. ADA4077 current path when V+ is absent (10 V input and 47 output load).
5.0 mA V+ 47 2.69 mA V+ 47
170 mA 6.27 mA
V+ V+
D3 V+ D3 V+
IN IN
8.01 V V+ D7 0.295 V V+ D7
D4 D4
D5 D6 OUT D5 D6 OUT
D1 15 V D8 6.06 V 9.94 V D1 15 V D8 1.876 V
+IN 170 mA 6.27 mA
D2 5.0 mA 15 V D2 15 V
175 mA 8.96 mA
ADA4077 5.0 mA ADA4077 2.69 mA
15 V 15 V
15 V 15 V
Figure 7. ADA4077 current path when V+ is absent (10 V input and 47 power load).
V+ V+
4.9 mA 1.6 mA
V+ 3.226 V V+
D3 0.3 mA V+ V+
D3
IN IN
V+ D7 10.83 V V+ D7
D4 OUT D4 OUT
D5 D6 0.3 mA 2.501 V D5 D6 0.7 mA
D1 15 V D8 48.3 mA D1 15 V D8 1 k
+IN 0.7 mA
+IN
53.2 mA
47 13.6 mA 12 mA 47
D2 4.6 mA 15 V D2 15 V
10.02 V
2.3 mA
9.97 V
15 V 15 V
Figure 8. ADA4077 current path when V+ is absent (10 V input and 47 output load).
V+ V+
2.63 mA 0.15 mA
V+ 3.2 V V+ 10.4 V
D3 0.16 mA V+ D3 0.15 mA V+
OVP IN OVP
V+ D4 D7 OUT V+ D4 D7 OUT 0.428 V
D5 D6 0.16 mA 2.543 V D5 D6 0.15 mA
9.97 D1 D8 10.00 D1 D8 1 k
+IN OVP 51.37 mA +IN OVP 8.95 mA
V 47 54 mA V 9.1 mA 47
D2 D2
2.47 mA
ADA4177 51.53 mA ADA4177 9.10 mA
V V
V V
Figure 9. ADA4177 current path when V+ is absent (10 V input and 47 output load).
V+ V+
51.9 mA
V+
V+
D3 V+
D3 V+
IN
V+ D7 IN
D4 8.71 V V+ D4 D7
D5 D6 OUT
D1 D8 D5 D6 8.2 mA OUT
9.56 V 8.2 mA D1 D8 7.99 V
+IN
D2 +IN
V
D2 V
ADA4084 60.1 mA ADA4084
V 60.1 mA
V
V
15 V
Figure 10. ADA4084 current path when V+ is absent (10 V input).
V+ absent Vo or V+ is floating and negative input 10.02 12.33 15 0 4.31 4.18 0.039 12.32
Table 9
ADA4084-2 Condition V+ V I+ (mA) I (mA) IB+ (mA) IOUT (mA) IN (V) OUT (V)
V+ absent and positive input 8.71 15 0 60.1 60.102 51.89 9.56 7.99
In system applications, different op amps, different topology (such as We tested three ADI op amps in a power supply absent application
noninverting amplification, inverting amplification, and difference amplifi- (ADA4084-2, ADA4077-2, and ADA4177-2). When integrated with internal
cation), different load, and external connections can be implemented. resistors, the ADA4077-2 proved to be very robust. The ADA4177, when
If one power supply is absent, the risks need to be evaluated. This article integrated with an OVP circuit, delivered the best robustness. In applications
can provide guidance on setting up the evaluation circuit (Figure 2), how where the power may be absent, and external current-limiting resistors cant
to analyze the current path, and evaluate the potential risks. be added, the ADA4177 is recommended to avoid degrading the precision.
Summary References
To avoid overvoltage or latch-up situations, operational amplifier power ADA4077 data sheet. Analog Devices, Inc.
supplies must be established simultaneously. General guidelines are:
ADA4177 data sheet. Analog Devices, Inc.
XX During the Power On sequence, turn on the supply first, then apply a
signal at the input Arkin, Michael and Eric Modica. Robust Amplifiers Provide Integrated
Overvoltage Protection. Analog Dialogue, Volume 46, Number 1, 2012.
XX During the Power Off, turn off the input signal first, then turn off the
power supply Blanchard, Paul and Brian Pelletier. Using ESD Diodes as Voltage Clamps.
In real-world applications, these guidelines may be difficult to adhere to. Analog Dialogue, Volume 49, Number 10, 2015.
This can cause problems, especially when there is an input signal, and
designers need to properly evaluate the risk. An effective solution is try to For more information on the ADA4177 and ADA4077, see the product
limit the input current of the op amp so it is within the specifications in the pages and data sheets here: ADA4177 and ADA4077.
data sheet. Adding a current-limiting resistor at the input and output can
help in applications where power cant be supplied at the same time.
David Guo
David Guo [david.guo@analog.com] is a product applications engineer for
ADIs linear products. He started working in the China Central Application Also by this Author:
Center of ADI as an applications engineer in 2007 and transferred to the Low Power, Unity-Gain
Precision Amplifier Group as an applications engineer in June, 2011. Since Difference Amplifier
Implements Low Cost
January 2013, David has worked as an application engineer in ADIs Linear
Current Source
Product Department, where he is responsible for the technical support of
products including precision amplifiers, instrumentation amplifiers, high Volume 45, Number 2
speed amplifiers, current sense amplifiers, multipliers, references, and
rms-to-dc products. David earned his bachelors and masters degree in
mechano-electronic engineering from Beijing Institute of Technology.
Share on
Introduction
Nondispersive infrared (NDIR) spectroscopy is often used to detect gas and The thermopile sensor is composed of a large number of thermocouples
measure the concentration of carbon oxides (for example, carbon monoxide connected usually in series or, less commonly, inparallel. The output
and carbon dioxide). An infrared beam passes through the sampling chamber, voltage of the series connected thermocouples depends on the temperature
and each gas component in the sample absorbs some particular frequency difference between the thermocouple junctions and the reference
of infrared. By measuring the amount of absorbed infrared at the appropriate junctions. This principle is called the Seebeck effect after its discoverer,
frequency, the concentration of the gas component can be determined. The Thomas Johann Seebeck.
technique is said to be nondispersive because the wavelength that passes The circuit uses the AD8629 op amp to amplify the thermopile sensor
through the sampling chamber is not prefiltered and instead the optical filter output signals. The relatively small output voltage of the thermopile (from
is in front of the detector to eliminate all light except the wavelength, which hundreds of microvolts to several millivolts) requires high gain with very
the selected gas molecules can absorb. low offset and drift to avoid dc errors. The high impedance (typically
84 k) of the thermopile requires low input bias current to minimize
The circuit shown in Figure 1 is a complete thermopile-based gas sensor
errors, and the AD8629 bias current is only 30 pA typical. The very low
using the NDIR principle. This circuit is optimized for CO2 sensing, but can
drift with time and temperature eliminates additional errors once the
also accurately measure the concentration of a large number of gases by
temperature measurement has been calibrated. A pulsed light source
using thermopiles with different optical filters.
synchronized with the ADC sampling rate minimizes the errors caused by
The printed circuit board (PCB) is designed in an Arduino shield form factor low frequency drift and flicker noise.
and interfaces to the EVAL-ADICUP360 Arduino-compatible platform board. The AD8629 only has 22 nV/Hz voltage noise spectral density at 1 kHz,
The signal conditioning is implemented with the AD8629 and the ADA4528-1 which is less than the thermopile voltage noise density of 37nV/Hz.
low noise amplifiers and the ADuCM360 precision analog microcontroller,
The AD8629 also has a very low current noise spectral density of 5fA/Hz
which contains programmable gain amplifiers, dual, 24-bit, - analog-to-
typical at 10 Hz. This current noise flows through the 84 k thermopile
digital converters (ADCs), and an ARM Cortex-M3 processor.
and only contributes 420 pV/Hz at 10Hz.
7 V to 12 V
EVAL-ADuCM360-ARDZ
ADC3
ADC0 ADP7105
15 nF
VIN
47 k VOUT
P1.3 EN/UVLO
220 3.3 V
Sense
AD8629 ADC1 To ADuCM360
Muxes, PGAs, ADCs
SS
8.2 nF GND
15 nF
C6
47 k 10 nF
220
AD8629 ADC2
3.3 V
8.2 nF Dual Thermopiles with
R3 Optical Filters and Thermistor Lamp
510 k
5V
R4 0V
130 k
CO2
3.3 V
200 mV
R7
5.1 k 3.3 V
CO2 CO2
R9
330
ADA4528-1
Figure 1. NDIR gas sensing circuit (simplified schematic: all connections and decoupling not shown) circuit description.
to drive the lamp, and is turned on and off by the ADuCM360. The soft start
feature of the ADP7105 eliminates the in-rush current when cold starting
the lamp. Reference Junctions at Reference Temperature
V
The ADuCM360 includes dual, 24-bit, - ADCs for simultaneous sampling
Figure 3. Thermopile constructed of multiple thermocouples.
of a dual element thermopile at programmable rates of 3.5Hz to 3.906 kHz.
The data rate in the NDIR system is limited from 3.5 Hz to 483 Hz for best In the NDIR application, pulsed and filtered IR light is applied to the series
noise performance. connected active junctions; the junctions are therefore heated, which in
turn generates a small thermoelectric voltage. The temperature of the ref-
Thermopile Detector Theory of Operation erence junction is measured with a thermistor.
To understand the thermopile, it is useful to review the basic theory
of thermocouples. Many gases have permanently or nonpermanently separated centers of
positive and negative charge. The gases are able to absorb specific fre-
If two dissimilar metals are joined at any temperature above absolute zero, quencies in the infrared spectrum, which can be used for gas analysis.
there is a potential difference between them (their thermoelectric EMF or When infrared radiation is incident on the gas, the energy states of atoms
contact potential), which is a function of the temperature of the junction vibrating in the molecules change in discrete steps when the wavelength
(see the thermoelectric EMF circuit in Figure 2). of the infrared matches the molecules natural frequencies or resonances.
If the two wires are joined at two places, two junctions are formed (see the For a majority of IR gas sensing applications, the identities of the target
thermocouple connected to load in Figure 2). Ifthe two junctions are at gases are known; therefore, there is little need for gas spectrometry. How-
different temperatures, there is a net EMF in the circuit, and a current flows ever, the application must deal with a certain amount of cross sensitivity
determined by the EMF and the total resistance in the circuit (see Figure 2). between different gases if their absorption lines overlap.
If one of the wires is broken, the voltage across the break is equal to the
net thermoelectric EMF of the circuit, and if this voltage is measured, it can Carbon dioxide has a very strong absorption band between 4200nm and
be used to calculate the temperature difference between the two junctions 4320 nm, as shown in Figure 4.
(see the thermocouple voltage measurement in Figure 2). Remember 1017
that a thermocouple measures the temperature difference between two
junctions, not the absolute temperature at one junction. The temperature at 1018
the measuring junction can be measured only if the temperature of the other
Absorption Intensity (cm/mol)
1019
junction (often called the reference junction or the cold junction) is known.
1020
However, it is not so easy to measure the voltage generated by a ther- 1021
mocouple. Suppose that a voltmeter is attached to the first thermocouple
measurement circuit (see the practical thermocouple voltage measurement 1022
showing cold junction in Figure 2). The wires attached to the voltmeter 1023
form further thermocouples where they are attached. If both these addi-
tional junctions are at the same temperature (it does not matter what 1024
temperature), the law of intermediate metals states that they make no net 1025
contribution to the total EMF of the system. If they are at different tem-
peratures, they introduce errors. Because every pair of dissimilar metals in 1026
1500 2000 2500 3000 3500 4000 4500 5000 5500
contact generates a thermoelectric EMFincluding copper/solder, kovar/ Wavelength (nm)
copper (kovar is the alloy used for IC lead frames), and aluminum/kovar Figure 4. Absorption spectrum of carbon dioxide (CO2).
(at the bond inside the IC)the problem is even more complex in practical
circuits, and it is necessary to take extreme care to ensure that all the The available output range of IR sources and the absorption spectrum
junction pairs in the circuitry around a thermocouple, except for the mea- of water also govern the choice of the sensing wavelengths. Water shows
surement and reference junctions themselves, are at the same temperature. strong absorptions below 3000nm and also between 4500 nm and
Copper Copper
Metal A Metal A R Metal A Metal A Metal A Metal A V Metal A
V1 V2
I T3 T4
Thermoelectric
V1 T1 EMF V1 T1 T2 V2 V1 T1 T2 V2 V1 T1 T2 V2
Metal B Metal B Metal B Metal B
R = Total Circuit Resistance
I = (V1 V2) R V = V1 V2, IF T3 = T4
1019
V is the output in target gas.
1020
1021
Rearranging and combining the previous two equations gives:
1022 FA = 1 e klx
1023
If k and l are held constant, FA can be plotted against x as shown in Figure 6
1024 (where kl = 115, 50, 25, 10, and 4.5). The value of FA increases with c, but
1025
eventually saturates at high gas concentrations.
1.2
1026
1500 2000 2500 3000 3500 4000 4500 5000 5500 FA = 1 eklx
Wavelength (nm)
1.0
Figure 5. Absorption spectrum of carbon dioxide overlaid with water.
kl = 115
Fractional Absorbance
0.8
If IR light is applied to a dual thermopile detector fitted with a pair of optical kl = 50
filters so that one filter is centered on 4260nm and the other on 3910 nm, kl = 25
the concentration of carbon dioxide can be measured from the ratios of the 0.6
two thermopile voltages. The optical filter that resides within the absorption kl = 10
channel serves as the detection channel and the optical filter that resides 0.4
outside the absorption spectrum serves as the reference channel. Measur-
ing errors caused by dust or diminishing radiation intensity are removed 0.2
by the use of the reference channel. It is important to note that there are kl = 4.5
no gas absorption lines at 3910nm, making this the ideal location for the 0
reference channel. 0 2 4 6 8 10
CO2 Volume Gas Concentration (%)
Thermopiles used in NDIR sensing have relatively high internal resistance Figure 6. Typical fractional absorbance for kl = 4.5, 10, 25, 50, 115.
and 50 Hz/60 Hz power-line noise can couple into the signal path. The ther-
mopile can have source impedances of about 100k causing the thermal This relationship implies that for any fixed setup, the ability to resolve a
noise to dominate the system. For example, the thermopile detector change in gas level is better at low concentrations than at high concentra-
chosen in the Figure 1 system has a voltage noise density of 37 nV/Hz. tions. However, k and l can be adjusted to provide the optimum absorbance
By maximizing the amount of signal coming from the detector and using for the required range of gas concentration. This means that long opti-
less gain in the circuit, it is possible to ensure the best performance of the cal paths are more suited for low gas concentrations, and short optical
gas measuring system. paths are more suited for high gas concentrations.
The best way to maximize the signal from the thermopile detector is to use The following describes a two-point calibration procedure necessary
a sample chamber with high reflective properties, which ensures that the to determine the kl constant using the ideal Beer-Lambert equation.
detector absorbs the radiation emitted from the source and not the cham- If b = kl, then:
ber itself. Using a reflective chamber to reduce the amount of radiation
absorbed by the chamber can also reduce the amount of power consumed I
by the system because a less powerful radiation source can be used. FA = 1
I0
ACT I LOW c
T ln REF ZERO 1 = SPAN 1 eb(xLOW )
x= I0
TLOW b
I CAL
1 = SPAN 1 eb(xCAL c)
I0
The T/TLOW factor compensates for the change in concentration with
temperature due to the ideal gas law. Solving the two equations for ZERO and SPAN yields:
I0 = ZERO =
Modified Beer-Lambert Law
c c
Practical considerations in the NDIR implementation require modifications ACTLOW eb(xCAL ) 1 REFCAL + ACTCAL 1 eb(xLOW ) REFLOW
to the Beer-Lambert Law, as follows, to obtain accurate readings:
b(xCAL c) b(xLOW c)
e e REFCAL REFLOW
c
FA = SPAN (1 e bx )
SPAN =
The SPAN factor is introduced because not all the IR radiation that impinges ACTCAL REFLOW ACTLOW REFCAL
upon the active thermopile is absorbed by the gas, even at high concentra-
c c
tions. SPAN is less than 1 because of the optical filter bandwidth and the ACTLOW eb(xCAL ) 1 REFCAL + ACTCAL 1 eb(xLOW ) REFLOW
fine structure of the absorption spectra.
Figure 7. Thermopile driver equivalent circuit, G = 214.6.
This equation assumes that TLOW = TCAL.
The step function setting time of the 84 k/8.2 nF filter to 22bits
Effects of Ambient Temperature is approximately
The thermopile detector senses temperature by absorbing radiation, but
= 84 k 8.2 nF ln222 10.5 ms
it also responds to ambient temperature changes that can give rise to
spurious and misleading signals. For this reason, many thermopiles have
thermistors integrated into the package. The AD8629 noninverting amplifier is set to a gain of 214.6 and the 3 dB
cutoff frequency:
The radiation absorption is related to the number of target molecules in the
1
chamber, not the absolute percentage of target gas. Therefore the absorp- f3dB 225.75 Hz
tion is described by the ideal gas law at standard atmosphere pressure. 2 47 k 15 nF
It is necessary to record the temperature data in both the calibration state The settling time to 22 bits is approximately
and the measurement state:
= 47 k 15 nF ln222 10.75 ms
T
xT = x
TLOW The maximum NDIR chop frequency is 5 Hz, and the minimum half cycle
pulse width is therefore 100 ms. The settling time to 22bits is approxi-
where: mately 0.1 the minimum chop pulse width.
x is the concentration of gas without temperature compensation.
TLOW is the temperature in K at low and high gas concentration. The AD8629 has a 0.1 Hz to 10 Hz input voltage noise of 0.5Vp-p. Ignor-
T is the temperature in K at sampling. ing the sensor voltage noise and the AD8629 current noise, a 1 mV p-p
xT is the gas concentration at temperature T. signal output from the thermopile yields a signal-to-noise ratio (SNR) of:
Thermopile Driver The gain of the AD8629 stage is 214.6, and the gain of the internal PGA
of ADuCM360 is automatically set by software from 1 to 128 to ensure
The HTS-E21-F3.91/F4.26 thermopile (Heimann Sensor, GmbH) has an
the input signal matches the full-scale span of the ADC input, 1.2 V.
84 k internal resistance in each channel. The equivalent circuit of the
The peak-to-peak signal from the thermopile can range from a few hun-
driver for one of the thermopile channels is shown in Figure 7. The internal
dred V to several mV. For instance, if the full-scale thermopile signal is
84 k thermopile resistance and the external 8.2 nF capacitor form an RC
1mVp-p, a PGA gain of 4 produces an 860mV p-p into the ADC.
low-pass noise filter with a 3dB cutoff frequency:
RTH = 100 k For the default chopping frequency of 0.25 Hz, the thermopile data is taken
at a 10 Hz rate during the last 1.5 sec of the 2sec half cycle to ensure that
= 3940
the signal has settled. The data during the first 500 ms is ignored (blanking
time). The blanking time can also be set in the software for both edges.
The Thevenin equivalent circuit for the thermistor driver is shown in Figure 8.
Note that the NTC thermistor data is taken during the blanking time.
The R3 and R4 divider resistors provide a 670.3mV voltage source in series
with the 103.6k. The driving voltage is 670.3mV 200mV = 470.3mV.
Calibration Procedure: Ideal Beer-Lambert Equation
ADC3 ADC0
Because of differences in the characteristics of lamps and thermopiles,
R3/R4 the circuit must be calibrated initially and also after changing either the
NTC 103.6 k
thermopile or the lamp.
200 mV 670.3 mV
It is recommended that the entire assembly be placed in a closed chamber
Figure 8. NTC thermistor driver equivalent circuit.
where gas of a known CO2 concentration can be injected until all existing
gas in the chamber is flushed out. After stabilizing for a few minutes, the
When RTH = 100 k at 25C, the voltage across the thermistor is 231mV, measurements can then be made.
and the PGA gain is therefore set at 4 when making the measurement.
The calibration method and algorithms are shown in the following steps for
The flexible input multiplexer and dual ADCs in the ADuCM360 allow the ideal Beer-Lambert equation:
simultaneous sampling of both the thermopile signals and the temperature
sensor signal to compensate for drift. 1. Input the following command: sbllcalibrate (standard Beer-Lambert
calibration).
IR Light Source Driver 2. Inject low concentration, xLOW, or zero gas (nitrogen), and stabilize
the chamber.
The filament light source selected is the International Light Technologies
MR3-1089, with a polished aluminum reflector that requires a drive voltage 3. Input the CO2 concentration into the terminal.
of 5.0 V at 150 mA for maximum infrared emission and the best system 4. The system measures ACTLOW, the peak-to-peak output of the active
performance. Heat from the lamp keeps the temperature of the optical detector in low concentration gas.
reflector higher than ambient, which is helpful in preventing condensation 5. The system measures REFLOW, the peak-to-peak output of the reference
in humid environments. detector in low concentration gas.
Filament lamps have a low resistance when cold (turned off), which can 6. The system measures temperature of low gas, TLOW.
result in a current surge at the instant of turn-on. A regulator with a soft 7. Inject high concentration CO2, of concentration xCAL, into the chamber.
start function is useful in addressing this problem. 8. Input the CO2 concentration into the terminal.
The ADP7105 low dropout voltage regulator has a programmable enable 9. The system measures ACTCAL, REFCAL, and the calibration
pin that can be used with a general-purpose input/output pin of the temperature, TCAL.
ADuCM360 to enable/disable the lamp voltage. A soft start capacitor, 10. The system calculates ZERO and b:
C6, of 10 nF provides a soft start time of 12.2 ms, which is approximately
xLOW
0.125 the minimum chop step time of 100ms.
ACTLOW ACTLOW REFCAL xCAL xLOW
ZERO =
The lamp on-current (~150 mA) is large, therefore careful circuit design REFLOW REFLOW ACTCAL
and layout is required to prevent the lamp switching pulses from coupling
into the small thermopile output voltages. ACTLOW REFCAL
ln
Take care to ensure the lamp return path does not flow through the sensitive REFLOW ACTCAL
thermopile detector ground return path. The lamp current must not use the b =
x CAL x LOW
same return path as the processorotherwise it may cause voltage offset
errors. It is strongly recommended that a separate voltage regulator be used
for the lamp drive and the signal conditioning portion of the system.
The ADP7105 lamp driver is supplied directly from the external power
supply connected to the EVAL-ADICUP360 board.
1. Apply the unknown concentration of gas to the chamber and stabilize. 1. Apply the unknown concentration of gas to the chamber and stabilize.
2. Measure ACT, the peak-to-peak output of the active detector. 2. Measure ACT, the peak-to-peak output of the active detector.
3. Measure REF, the peak-to-peak output of the reference detector. 3. Measure REF, the peak-to-peak output of the reference detector.
4. Measure the temperature, T, in Kelvin. 4. Measure the temperature, T, in kelvin.
5. Use the ZERO value from the calibration. 5. Use the ZERO and SPAN values from calibration.
6. Use the b value from the calibration. 6. Use the values of b and c that were previously determined.
7. Calculate the fractional absorbance: 7. Calculate fractional absorbance:
ACT
FA = 1 ACT
REF ZERO FA = 1
REF ZERO
Calculate the concentration and apply the ideal gas law temperature
Calculate the concentration and apply the ideal gas law temperature
compensation:
compensation:
ACT 1
T ln REF ZERO
x= FA c
TLOW b T ln 1 SPAN
x=
TLOW b
This procedure assumes that TLOW = TCAL.
This procedure assumes that TLOW = TCAL.
Note that the CN-0338 software will automatically perform Steps 2
through 7.
NTC Thermistor Algorithm and Calculations
Calibration Procedure: Modified Beer-Lambert Equation The NTC thermistor equivalent circuit is shown in Figure 9.
If the constants b and c are known from measurements, use the following 3.3 V
procedure. ADC3 3.3 V ADC0
R7
5.1 k R3
200 mV
1. Input the following command: mbllcalibrate (modified Beer-Lambert NTC 510 k
calibration). R9
330 R4
2. Input the b and c constants. ADA4528-1
130 k
T=
T0
R3 R4 (R7 + R9) VNTC
+ T0 ln
(R4 R7 R3 R9) RTH VCC (R3 + R4) (R7 + R9) RTH VNTC
During each lamp chopping time interval, the ADC is switched to NTC A functional diagram of the test setup is shown in Figure 12 and a
sampling, as shown in Figure 10. photograph of the EVAL-CN0338-ARDZ Arduino shield board and the EVAL-
ADICUP360 Arduino-compatible platform board is shown in Figure 13.
NTC NTC NTC
Sample Sample Sample
7 V to 12 V EVAL-CN0338-ARDZ PC
Power
Thermopile Thermopile Thermopile
Sample Sample Sample
USB
EVAL-ADICUP360
Figure 11 shows the fractional absorbance (FA) as a function of CO2 SPI Pmod
concentration for a typical EVAL-CN0338-ARDZ board. I2C Pmod
0.20
EVAL-ADICUP360
0.18 Arduino-Compatible
Development
Platform Board
0.16
Fractional Absorbance
0.14
0.12
User USB
0.10
Debug USB
0.08 Power
Figure 13. EVAL-CN0338-ARDZ board and EVAL-ADICUP360
0.06
board photos.
0.04
0.02 Summary
0 The analog electronics needed to implement the NDIR measurement
require precision low noise amplification and high resolution analog-
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
50000
Application Note2, Signal Processing for Infrared Gas Sensors. HITRAN Catalog.
SGX Sensortech, 2007.
Micro-Hybrid, Construction and Function of Thermopiles.
Application Note 3, Design of Microcontroller Software for Infrared Gas
Sensors. SGX Sensortech, 2007. MT-004 Tutorial, The Good, the Bad, and the Ugly Aspects of ADC Input
NoiseIs No Noise Good Noise? Analog Devices, Inc., 2009.
Application Note 4, Design of Electronics for Infrared Gas Sensors.
SGX Sensortech, 2009. MT-031 Tutorial, Grounding Data Converters and Solving the Mystery of
AGND and DGND. Analog Devices, Inc., 2009.
Application Note 5, Determining Coefficients for Linearisation and
Temperature Compensation. SGX Sensortech, 2009. MT-035, Op Amp Inputs, Outputs, Single-Supply, and Rail-to-Rail Issues.
Analog Devices, Inc., 2009.
Application Note AAN-201, NDIR: Gas Concentration Calculation Overview.
Alphasense Limited, 2014. MT-037 Tutorial, Op Amp Input Offset Voltage. Analog Devices, Inc., 2009.
Application Note AAN-202, NDIR: Electronic Interface and Signal Extraction MT-101 Tutorial, Decoupling Techniques. Analog Devices, Inc., 2009.
for Pyroelectric Sensor. Alphasense Limited, 2016.
Data Sheets and Evaluation Boards
Application Note AAN-203, NDIR: Determination of Linearisation and ADP7105 data sheet.
Temperature Correction Coefficients. Alphasense Limited, 2009.
ADuCM360 data sheet.
Application Note AAN-204, NDIR: Origin of Nonlinearity and SPAN.
Alphasense Limited, 2009. EVAL-CN0338-ARDZ Arduino Shield Board.
Application Note AAN-205, NDIR: Running the IRC-A1 at Reduced Lamp EVAL-ADICUP360 Arduino-Compatible Platform Board.
Voltages. Alphasense Limited, 2009.
Robert Lee
Robert Lee [robert.lee@analog.com] has been an applications engineer
at Analog Devices since January 2013. Robert received his B.S.E.E. from
University of Electronic Science and Technology of China (UESTC) in 2004
and M.S.E.E. from UESTC in 2009. He has more than 10 years of embedded
system design experience.
Walt Kester
Walt Kester [walt.kester@analog.com] is a corporate staff applications
engineer at Analog Devices. During his many years at ADI, he has designed,
developed, and given applications support for high speed ADCs, DACs, SHAs,
op amps, and analog multiplexers. An author of many papers and articles,
he prepared and edited 11 major applications books for ADIs global technical
seminar series; topics include op amps, data conversion, power management,
sensor signal conditioning, mixed-signal circuits, and practical analog design
techniques. His latest book, Data Conversion Handbook (Newnes), is a nearly
1000-page comprehensive guide to data conversion. Walt has a B.S.E.E. from
NC State University and an M.S.E.E. from Duke University.
Share on
tricky. Take a look at Figure 1, which shows the block diagram schematic
of the AD9680-1250 being clocked at 1.25 GHz. Everything here looks
normal to someone taking a first look at the schematic. The decoupling
capacitors are not shown here, as neither are other supply domains. The
focus is on the 1.25 V domain as this is the smallest supply voltage.
1.25 V 1.12 V
~1.3 A
ADP1741
AD9680
Question:
CLK
In my system the SPI interface for the ADC is returning 0xFF for every read.
What could be wrong? 1.25 GHz
4. Split the voltage outputs to the respective domains (AVDD1, AVDD1_SR, So the next time you plug in an ADC, clock it, and find that it doesnt
DVDD, DRVDD) and use a ferrite bead with lower DCR to ensure proper work and the SPI is returning 0xFF for every read cycle, you might have
operating voltage. Mr. Ohm to thank. In this case, the venerable DMM becomes your tool of
choice, not an oscilloscope or your friendly applications engineer. A ferrite
Figure 2 shows Options 2 and 4, discussed above. Option 4 offers the bead certainly can offer good noise immunity to your system. However, if
best compromise. However, this does add to the bill of material (BOM) not chosen correctly, and if Ohms law is not heeded, this small component
cost, which must be taken into consideration. Option 4 also provides can cause some serious issues in realizing an ADCs true performance in
the system.
1.275 V
1.25 V 1.12 V
~1.3 A
ADP1741
~1.3 A
ADP1741
Ferrite Bead
DCR = 30 m
Ferrite Bead
DCR = 100 m 802 mA 250 mA 220 mA
802 mA 250 mA 220 mA
AD9680 AD9680
CLK CLK
Figure 2: Different options of choosing and using a ferrite bead to power the AD9680.
Umesh Jayahoman
Umesh Jayamohan [umesh.jayamohan@analog.com] is an applications
engineer with Analog Devices in the High Speed Converter Group Also by this Author:
(Greensboro, NC). He has been a part of Analog Devices since 2010. Rarely Asked
Umesh received his B.S.E.E. from the University of Kerala, India, in 1998, Questions
and his M.S.E.E. from Arizona State University in 2002. Issue 129, May 2016
Who Ate My dBs?
Share on
In the first part of this article, Whats Up with Digital Downconverters We will begin by looking at the DDC decimation filters when the complex
Part 1, we looked at the industry push for sampling higher frequencies in to real conversion block is enabled in the AD9680. This means the DDC will
higher frequency RF bands and how digital downconverters (DDCs) can be configured to accept a real input and have a real output. In the AD9680,
enable this type of radio architecture. Several technical aspects were dis- the complex to real conversion automatically shifts the input frequencies
cussed relating to the DDC that resides in the AD9680 family of products. up in frequency by an amount equal to fS/4. Figure 2 shows the low-pass
One such aspect was that higher input sampling bandwidths allow for response of the HB1 filter. This is the response of HB1 showing the real
radio architectures that can directly sample at higher RF frequencies and and complex domain response. In order to understand the real operation
convert the input signals directly to baseband. The DDC enables an RF of the filter it is important to first see the basic filter response in the real
sampling ADC to digitize such signals without the expense of a large amount and complex domains so that the low-pass response can be seen. The HB1
of data throughput. The tuning and decimation filtering that resides in the filter has a pass band of 38.5% of the real Nyquist zone. It also has a stop
DDC can be utilized to tune the input band and filter undesired frequencies. band that is 38.5% of the real Nyquist zone with the transition band making
In this installment we will take a closer look at the decimation filtering and up the remaining 23%. Likewise in the complex domain, the pass band and
apply it to the example that was discussed in Part 1. In addition we will stop band each make up 38.5% (77% total) of the complex Nyquist zone
take a look at Virtual Eval, which incorporates the ADIsimADC engine into with the transition band making up the remaining 23%. As Figure 2 illus-
a new and revamped software simulation tool. Virtual Eval will be used to trates, the filter is a mirror image between the real and complex domains.
demonstrate how closely the simulated result matches the measured data
from the example.
the signal first passes through the NCO, which shifts the input tones in Figure 2. HB1 filter responsereal and complex domain response.
frequency, then passes through the decimation, optionally through the
gain block, and optionally through the complex to real conversion. Now we can observe what happens when we place the DDC into real mode
by enabling the complex to real conversion block. Enabling the complex to
DDC 0 real conversion results in a shift of fS/4 in the frequency domain. This is illus-
Real/I Real/I trated in Figure 3, which shows the frequency shift and the resulting filter
DCM = Bypass or 2
DCM = Bypass or 2
DCM = Bypass or 2
GAIN = 0 or 6 dB
Complex to Real
Converter 0 response. Notice the solid lines and the dotted lines of the filter response.
Conversion
(Optional)
NCO
HB21 FIR
DCM = 2
HB4 FIR
HB3 FIR
HB2 FIR
+
Mixer
The solid line and shaded area indicates this is the new filter response
Real/Q (Optional) Real/Q after the fS/4 frequency shift (the resulting filter response cannot cross the
Converter 1 Nyquist boundary). The dotted lines are given for illustration to show the
filter response that would exist if not for running into the Nyquist boundary.
Figure 1. DDC signal processing blocks in the AD9680.
Now lets look at the case where we enable HB1 + HB2. This results in
a decimation ratio of two. Once again, the actual frequency response of
100 dB the HB1 + HB2 filters is given by the solid blue line. The center frequency
of the filter pass band is still fS/4. Enabling both HB1 + HB2 filters results
in an available bandwidth of 38.5% of the Nyquist zone. Once again,
notice the aliasing effects of the ADC and its impact on the combination
fS/4 0 fS/4 fS/2 3fS/2 of HB1 + HB2 filters. A signal that appears at 7fS/8 will alias into the first
Figure 3. HB1 filter responsereal DDC mode (complex to real Nyquist zone at fS/8. Likewise a signal at 5fS/8 will alias into the first
conversion enabled). Nyquist zone at 3fS/8. These examples with the complex to real conver-
sion block enabled can easily be extended from HB1 + HB2 to include
Notice that the HB1 filter bandwidth remains unchanged between one or both of the HB3 and HB4 filters. Note that the HB1 filter is nonby-
Figures 2 and 3. The difference between the two is the fS/4 frequency passable when the DDC is enabled while HB2, HB3, and HB4 filters can
shift and the resultant center frequency within the first Nyquist zone. Notice optionally be enabled.
however that in Figure 2 we have 38.5% of Nyquist for the real portion of
the signal and 38.5% of Nyquist for the complex portion of the signal. In
Figure 3 with the complex to real conversion block enabled there is 77% 38.5% fS/2
<0.001 dB
of Nyquist for the real signal and the complex domain has been discarded. Actual Filter Response
Aliased Filter Response
The filter response remains unchanged apart from the fS/4 frequency shift.
Also, notice as a product of this conversion the decimation rate is now
equal to one. The effective sample rate is still fS but instead of the entire
Nyquist zone there is only 77% of available bandwidth in the Nyquist zone.
This means that with the HB1 filter and the complex to real conversion 100 dB
block enabled the decimation rate equals one (see the AD9680 data fS = Input Sample Clock
Next we will look at the filter responses of different decimation rates (that Figure 5. HB1 + HB2 effective filter response due to ADC aliasing
is, enabling multiple half-band filters) and how aliasing of the ADC input (decimation rate = 2).
frequencies impacts the effective decimation filter responses. The actual
frequency response of HB1 is given by the solid blue line in Figure 4. The Now that the real mode operation with the decimation filters enabled has
dashed line represents the effective aliased response of HB1 due to the been discussed, the complex mode of operation with the DDC can now be
aliasing effects of the ADC. Due to the fact that frequencies input into 2nd, examined. The AD9680 will continue to be used as an example. Similar
3rd, 4th, etc. Nyquist zones alias into the 1st Nyquist zone of the ADC, to the real mode operation of the DDC, the normalized decimation filter
the HB1 filter response is effectively aliased into these Nyquist zones. responses will be presented. Once again, the example filter response plots
For example, a signal residing at 3fS/4 will alias into the first Nyquist zone included here do not show the specific insertion loss vs. frequency, but
at fS/4. It is important to understand that the HB1 filter response resides instead they figuratively show the approximate response of the filter. This
only in the first Nyquist zone and that it is the aliasing of the ADC that is done to give a high level understanding of how the filter responses are
results in the effective response of the HB1 filter appearing to be aliased affected by the ADC aliasing.
into the other Nyquist zones.
This discussion of the DDC operation has given a good insight into both
38.5% fS/2 the real and complex modes of operation of the decimation filters in the
>100 dB
AD9680. There are several advantages that are offered by utilizing the
100 dB
fS = Input Sample Clock
decimation filtering. The DDC can operate in real or complex mode and
allow the user to use different receiver topologies depending on the needs
fS 3fS/4 fS/2 fS/4 0 fS/4 fS/2 3fS/4 fS of the particular application. This can now be put together with what was
Figure 6. HB1 effective filter response due to ADC aliasing
discussed in Part 1 and help to look at a real example with the AD9680.
(decimation rate = 2)complex. This example will put measured data together with simulated data from
Virtual Eval so that the results can be compared.
38.5% fS/4
<0.001 dB
38.5% fS/4
>100 dB
100 dB
fS = Input Sample Clock
fS 7fS/8 3fS/4 5fS/8 fS/2 3fS/8 fS/4 fS/8 0 fS/8 fS/4 3fS/8 fS/2 5fS/8 3fS/4 7fS/8 fS
Figure 7. HB1 + HB2 effective filter response due to ADC aliasing (decimation rate = 4)complex.
fs/2 7fs/16 3fs/8 5fs/16 fs/4 3fs/16 fs/8 fs/16 fs/16 fs/8 3fs/16 fs/4 5fs/16 3fs/8 7fs/16 fs/2
245.76 245.76
fs/2 7fs/16 3fs/8 5fs/16 fs/4 3fs/16 fs/8 fs/16 fs/16 fs/8 3fs/16 fs/4 5fs/16 3fs/8 7fs/16 fs/2
4.94 MHz
Spectrum After Decimate by 2 HB1 Filter
3rdHarmonic and Image Reduced 2nd Harmonic
fs/2 7fs/16 3fs/8 5fs/16 fs/4 3fs/16 fs/8 fs/16 fs/16 fs/8 3fs/16 fs/4 5fs/16 3fs/8 7fs/16 fs/2
4.94 MHz
Spectrum After Decimate by 4 HB2 Filter
3rd Harmonic Filtered
36.38 MHz
2nd Harmonic
59.28 MHz
Image
fs/2 7fs/16 3fs/8 5fs/16 fs/4 3fs/16 fs/8 fs/16 fs/16 fs/8 3fs/16 fs/4 5fs/16 3fs/8 7fs/16 fs/2
Figure 8. Signals as they pass through the DDC signal processing blockdecimation filtering shown.
In this example the same conditions as were used in Part 1 will be Spectrum After Decimate by 2:
used. The input sample rate is 491.52 MSPS and the input frequency is 1. The fundamental frequency stays at 4.94 MHz.
150.1 MHz. The NCO frequency is 155 MHz and the decimation rate is set
to four (due to the NCO resolution, the actual NCO frequency is 154.94 MHz). 2. The image of the fundamental translates down to 59.28 MHz and is
This results in an output sample rate of 122.88 MSPS. Since the DDC is attenuated by the HB1 decimation filter.
performing complex mixing the complex frequency domain is included in
the analysis. Note that the decimation filter responses have been added 3. The 2nd harmonic stays at 36.38 MHz.
and are shown in dark purple in Figure 8.
4. The 3rd harmonic is attenuated by the HB1 decimation filter.
Spectrum After the NCO Shift:
Spectrum After Decimate by 4:
1. The fundamental frequency shifts from +150.1 MHz down to 4.94 MHz.
1. The fundamental stays at 4.94 MHz.
2. The image of the fundamental shifts from 150.1 MHz and wraps
around to +186.48 MHz. 2. The image of the fundamental stays at 59.28 MHz and is attenuated
by the HB2 decimation filter.
3. The 2nd harmonic shifts from 191.32 MHz down to 36.38 MHz.
3. The 2nd harmonic stays at 36.38 MHz and is attenuated by the HB2
4. The 3 harmonic shifts from +41.22 MHz down to 113.72 MHz.
rd
decimation filter.
4.94 MHz
59.28 MHz
Figure 11. AD9680 speed grade selection and block diagram in
Virtual Eval.
36.38 MHz
Next, the input conditions must be set in order to perform the FFT simu-
lation (see Figure 12). Recall the test conditions for the example include a
clock rate of 491.52 MHz and an input frequency of 150 MHz. The DDC is
enabled with the NCO frequency set to 155 MHz, the ADC input is set to
Real, the complex to real conversion (C2R) is Disabled, the DDC decimation
rate is set to Four, and the 6 dB gain in the DDC is Enabled. This means
the DDC is set up for a real input signal and a complex output signal with
a decimation ratio of four. The 6 dB gain in the DDC is enabled in order
Figure 9. FFT complex output of signal after DDC with NCO = 155 MHz to compensate for the 6 dB loss due to the mixing process in the DDC.
and decimate by 4.
Virtual Eval will show only noise or distortion results at a time, so two plots
Now Virtual Eval can be used to see how the simulated results compare are included where one shows the noise results (Figure 12) and the other
to the measured results. To begin, open the tool from the website and shows the distortion results (Figure 13).
select an ADC to simulate (see Figure 10). The Virtual Eval tool is on the
Analog Devices website at Virtual Eval. The AD9680 model that resides in
Virtual Eval incorporates a new feature being developed that allows the
user to simulate different speed grades of ADCs. This feature is key to the
example since the example utilizes the AD9680-500. Once Virtual Eval
loads, the first prompt is to select a product category and a product. Notice
that Virtual Eval not only covers high speed ADCs but also has product
categories for precision ADCs, high speed DACs, and integrated/special
purpose converters.
Select the AD9680 from the product selection. This will open up the main
page for the simulation of the AD9680. The Virtual Eval model for the
AD9680 also includes a block diagram that gives details on the internal
configuration of the ADC analog and digital features. This block diagram
is the same as the one given in the data sheet for the AD9680. From this
page select the desired speed grade from the drop-down menu on the
left side of the page. For the example here, select the 500 MHz speed
grade as shown in Figure 11.
Figure 13. AD9680 FFT simulation in Virtual Evaldistortion results.
Jonathan Harris
Jonathan Harris [ jonathan.harris@analog.com] is a product applications
engineer inthe Space Products Group at Analog Devices in Greensboro, NC.
He has over 10 years of experience as an applications engineer support-
ing products in the RF industry. Jonathan received his M.S.E.E. from Auburn
University and his B.S.E.E. from UNC Charlotte. In his spare time he enjoys
motorcycle riding, college football, mobile audio, and spending time with
his family.
Share on
Introduction
The increasing importance of monitoring water quality has led to the devel-
opment of a number of related sensors and signal conditioning circuits.
Water quality is measured in terms of bacteria count, pH level, chemical
content, turbidity, and conductivity. All aqueous solutions conduct electricity
To I/V
to some degree. Adding electrolytes such as salts, acids, or bases to pure Converter
V
water increases the conductivity and decreases resistivity. This article
focuses on conductivity measurements.
EVAL-CN0359-EB1Z
Pure water does not contain significant amounts of electrolytes and
conducts only a small amount of electric current when a sample is sub- A = Area of Electrode Surface (cm2)
L = Distance Between Electrodes (cm)
jected to an applied voltageits conductivity is therefore low. On the other A I A V = Excitation Voltage
hand, a large quantity of electrolytes in the sample causes more current to I = Cell Current
K = L/A = Cell Constant (cm1)
be conductedits conductivity is higher. Y = Measured Conductance = I/V (S)
L YX = Water Conductivity = K Y (S/cm)
It is more common to think in terms of resistance rather than conduc-
tance, but the two are reciprocals. The resistivity, , of a material or
liquid is defined as the resistance of a cube of the material with perfectly Figure 1. Interface between conductivity cell and electronics
conductive contacts on opposite faces. The resistance, R, for other (EVAL-CN0359-EB1Z).
shapes, can be calculated by
The electronic circuitry impresses an alternating voltage on the sensor
R = L/A (1) and measures the size of the resulting current, which is related to the
conductivity. Because conductivity has a large temperature coefficient
where: (up to 4%/C), an integral temperature sensor is incorporated into the
Lis the distance between the contacts. circuitry to adjust the reading to a standard temperature, usually 25C
Ais the area of the contacts. (77F). When measuring solutions, the temperature coefficient of the
conductivity of the water itself must be considered. To compensate accu-
Resistivity is measured in units of cm. A 1 cm material has a resistance rately for the temperature, a second temperature sensor and compensation
of 1 when contacted on opposite faces of a 1 cm 1 cm 1 cm cube. network must be used.
Conductance is simply the reciprocal of resistance, and conductivity is the The contacting type sensor typically consists of two electrodes that are
reciprocal of resistivity. The unit of measurement of conductance is siemens insulated from one another. The electrodes, typically Type 316 stainless
(S), and the unit of measurement of conductivity is S/cm, mS/cm, or S/cm. steel, titanium palladium alloy, or graphite, are specifically sized and
spaced to provide a known cell constant. Theoretically, a cell constant of
For the purposes of this article, Y is the general symbol for conductivity
1.0/cm describes two electrodes, each sized 1 cm2in area, and spaced
measured in S/cm, mS/cm, or S/cm. However, in many cases, the
1 cm apart. Cell constants must be matched to the measurement system
distance term is dropped for convenience, and the conductivity is simply
for a given range of operation. For instance, if a sensor with a cell
expressed as S, mS, or S.
constant of 1.0/cm is used in pure water with a conductivity of 1 S/cm,
the cell has a resistance of 1 M. Conversely, the same cell in seawater
Measuring Conductivity Using Conductivity Cells has a resistance of 30 . Because the resistance ratio is so large, it is
A conductivity system measures conductivity by means of electronics difficult for ordinary instruments to accurately measure such extremes
connected to a sensor, called a conductivity cell, immersed in a solution, with only one cell constant.
as shown in Figure 1.
The cell constant, K, is defined as the ratio of the distance between the Take special care with sensors that have shields, as in the case of coaxial
electrodes, L, to the area of the electrodes, A: sensors. The shield must be connected to the same potential as the metal
container holding the liquid. If the container is grounded, the shield must
K = L/A (2) be connected to the circuit board ground.
The instrumentation then measures the cell conductance, Y: The final precaution is not to exceed the rated excitation voltage or
current for the cell. The following circuit allows programmable excitation
Y = I/V (3) voltages from 100 mV to 10 V, and the R23 (1 k) series resistor limits
the maximum cell current to 10 mA.
The conductivity of the liquid, YX, is then calculated:
1 nF AD8542 1 nF
IP-P R38
1 nF 1.2 k
ADG1211
R41 820 pF
R47,1.0 k
G2 = 1, 10, 100, 1000 430 +15 V R37
+15 V ADA4638-1 7.5 k +3.3 V
+15 V AD8253 C84 U16 24
A0, A1, 4.7 F
PWM1 U20B ADC1
ADA4627-1 U19 U4WRI
V2P-P 15 V
U18 +1.65 V ADA4528-2 10 nF
430 1 k
15 V REF R52
ADG1211 1 nF 1.2 k
VOUT2 =
15 V R42 G1 0.16 V2P-P
430 +15 V R48
G2 VOUT2 820 pF
YX = 1 mS ADA4638-1 7.5 k
G1 VOUT1 +1.65 V U22A C93
4.7 F U21 24
PWM2 U20A ADC0
15 V
+1.65 V ADA4528-2 10 nF
1 k
AD8542 1 nF
Figure 3. High performance conductivity measurement system (simplified schematic: all connections and decoupling not shown).
The +VEXC and VEXC voltages are generated by theADA4077-2op amps a dropout voltage of less than 1.2 V. The U9A op amp has a closed-loop gain
(U9A and U9B), and their amplitudes are controlled by the DAC output of of 8.33 and converts the ADuCM360 internal DAC output (0 V to 1.2 V) to
the ADuCM360, as shown in Figure 4. the +VEXC voltage of 0 V to 10 V. The U9B op amp inverts the +VEXC and
generates the VEXC voltage. R22 is chosen such that R22 = R24||R27 to
R14
110 k achieve first-order bias current cancellation. The error due to the 15 V
offset voltage of U9A is approximately (2 15 V) 10 V = 3 ppm. The
1 nF
R15
primary error introduced by the inverting stage is therefore the error in the
15 k +15 V resistor matching between R24 and R27.
G = 8.33 +VEXC
R16
DAC 13 k
U9A The ADG1419 is a 2.1 , on-resistance SPDT analog switch with an
from + ADA4077-2 0 V to +10 V
ADuCM360 on-resistance flatness of 50 m over a 10 V range, making it ideal
0 V to +1.2 V
R27
for generating a symmetrical square wave from the VEXC voltages.
27 k The symmetry error introduced by the ADG1419 is typically 50 m
1 nF
1 k = 50 ppm. Resistor R23 limits the maximum current through the
R24 sensor to 10 V/1 k = 10 mA.
27 k
VEXC The voltage applied to the cell, V1, is measured with the AD8253
R22 U9B
13 k
0 V to 10 V
instrumentation amplifier (U15). The positive input to U15 is buffered by
+ ADA4077-2
15 V
the ADA4000-1 (U14). The ADA4000-1 is chosen because of its low bias
current of 5 pA to minimize the error in measuring low currents associ-
Figure 4. Excitation voltage sources.
ated with low conductivities. The negative input of the AD8253 does not
The ADA4077-2 has a typical offset voltage of 15 V (A grade), a 0.4 nA bias require buffering.
current, a 0.1 nA offset current, and an output current of up to 10 mA, with
The offset voltages of U14 and U15 are removed by the synchronous
sampling stage and do not affect the measurement accuracy.
P-P (6)
Cell Voltage +VEXC
0
The cell current is given by
VEXC
P-P P-P
(7)
PWM0
Track +V
Solving Equation 8 for IP-P and substituting into Equation 7 yields the
PWM2
Hold +V Hold +V following for YX:
tSETUP tHOLD V2P-P
YX = (9)
Figure 5. Cell voltage and track-and-hold timing signals. V1P-P R 47
The output of the AD8253 in amp (U15) drives two parallel track-and-hold
circuits composed of ADG1211 switches (U17A/U17B), series resistors Solving Equation 5 and Equation 6 for V1P-P and V2P-P and substituting into
(R34/R36), hold capacitors (C50/C73), and unity-gain buffers (U10/U13). Equation 9 yields the following:
G2 VOUT2
The ADG1211 is a low charge injection, quad SPST analog switch, operat- YX = (10)
G1 VOUT1 R 47
ing on a 15 V power supply with up to 10 V input signals. The maximum
charge injection due to switching is 4 pC, which produces a voltage error G2 VOUT2
of only 4 pC 4.7 F = 0.9 V. YX = 1 mS (11)
G1 VOUT1
The PWM1 signal causes the U10 track-and-hold buffer to track the negative Equation 11 shows that the conductivity measurement depends on G1, G2,
cycle of the sensor voltage and then hold it until the next track cycle. The output and R47, and the ratio of VOUT2 to VOUT1. Therefore, a precision reference
of the U10 track-and-hold buffer is therefore a dc level corresponding to is not required for the ADCs within the ADuCM360.
the negative amplitude of the sensor voltage square wave.
The AD8253 gain error (G1 and G2) is 0.04% maximum, and R47 is chosen
Similarly, the PWM2 signal causes the U13 track-and-hold buffer to track to be a 0.1% tolerance resistor.
the positive cycle of the sensor voltage and then hold it until the next track
cycle. The output of the U13 track-and-hold buffer is therefore a dc level From this point, the resistors in the VOUT1 and VOUT2 signal chain determine
corresponding to the positive amplitude of the sensor voltage square wave. the overall system accuracy.
The bias current of the track-and-hold buffers (ADA4638-1) is 45 pA The software sets the gain of each AD8253 as follows:
typical, and the leakage current of the ADG1211 switch is 20 pA typical.
Therefore, the worst-case leakage current on the 4.7 F hold capacitors XX If the ADC code is over 94% of full scale, the gain of the AD8253 is
is 65 pA. For a 100 Hz excitation frequency, the period is 10 ms. The drop reduced by a factor of 10 on the next sample.
voltage over one-half the period (5 ms) due to the 65 pA leakage current XX If the ADC code is less than 8.8% of full scale, the gain of the AD8253
is (65 pA 5 ms) 4.7 F = 0.07 V. is increased by a factor of 10 on the next sample.
The following five resistors affect the accuracy in the VOUT2 current Figure 7 shows the configuration for 4-wire RTDs.
channel: R47, R37, R38, R48, and R52. ADuCM360
R13
1.5 k AIN7/
J3
Assuming that all nine resistors are 0.1% tolerance, and including the 0.04% RP 1 0.1% IEXC IEXC V7 V8
gain error of the AD8253, a worst-case error analysis yields approximately 100 or AIN8
to ADC
1000
0.6%. The analysis is included in the CN-0359 Design Support Package. Pt RTD RP 2 AIN6
+3.3 V
Accuracy measurements were taken using precision resistors from 1 to
1 k
1 M (1 S to 1 S) to simulate the conductivity cell. Figure 6 shows the RP 4 +0.115 V AIN9
results, and the maximum error is less than 0.1%.
36
0.04 V6 V5
RX = 1.5 k
V7 V8
0 The parasitic resistance in each of the leads to the remote RTD is shown
as RP. The excitation current (IEXC) passes through a precision 1.5 k
0.02 resistor and the RTD. The on-chip ADC measures the voltage across the
Error (%)
RTD (V6 V5) and uses the voltage across R13 (V7 V8) as a reference.
0.04
It is important that the R13 resistor and the IEXC excitation current value
0.06 be chosen such that the ADuCM360 maximum input voltage at AIN7 does
not exceed AVDD 1.1 V; otherwise, the IEXC current source does not
0.08 function properly.
0.10 The RTD voltage is accurately measured using the two sense leads that
1 10 0.1m 1m 10m 0.1 1
Conductivity (S)
connect to AIN6 and AIN5. The input impedance is approximately 2 M
(unbuffered mode, PGA gain = 1), and the current flowing through the
Figure 6. System error (%) vs. conductivity of 1 S to 1 S.
sense lead resistance produces minimum error. The ADC then measures
the RTD voltage (V6 V5).
RTD Measurement
Conductivity measuring system accuracy is only as good as its temperature The RTD resistance is then calculated as:
compensation. Because common solution temperature coefficients vary in
V6 V5
the order of 1%/C to 3%/C or more, measuring instruments with adjustable RX = 1.5 k (12)
temperature compensation must be used. Solution temperature coefficients V7 V8
are somewhat nonlinear and usually vary with the actual conductivity as
well. Therefore, calibration at the actual measuring temperature yields the The measurement is ratiometric and does not depend on an accurate
best accuracy. external reference voltage, only the tolerance of the 1.5 k resistor. In
addition, the 4-wire configuration eliminates the error associated with
The ADuCM360 contains two matched, software configurable, excitation the lead resistances.
current sources. They are individually configurable to provide a current
output of 10 A to 1 mA, and matching is better than 0.5%. The current The ADuCM360 has a buffered or unbuffered input option. If the internal buffer
sources allow the ADuCM360 to easily perform 2-wire, 3-wire, or 4-wire is activated, the input voltage must be greater than 100 mV. The 1 k/36
measurements for either Pt100 or Pt1000 RTDs. The software also resistor divider provides a 115 mV bias voltage to the RTD that allows buff-
automatically detects if the RTD is Pt100 or Pt1000. ered operation. In the unbuffered mode, Terminal 4 of J3 can be grounded
and connected to a grounded shield for noise reduction.
36 47 k
V8 V5
RX = 1.5 k
V7 V8 Figure 10. Power supply circuits.
Figure 8. Configuration for 3-wire RTD connection.
Figure 11 shows the LCD backlight driver circuit.
The second matched IEXC current source (AIN5/IEXC) develops a voltage
3.3 V
across the lead resistance in series with Terminal 3 that cancels the
100 nF
voltage dropped across the lead resistance in series with Terminal 1. The 5.1 k
2.9 V
measured V8 V5 voltage is therefore free of lead resistance error.
3.3 V 36 k 3.3 V
Figure 9 shows the 2-wire RTD configuration where there is no compensa-
tion for lead resistance. K1 K2
3.3 V 6.8 U6B U6A 6.8 3.3 V
R13 ADuCM360
1.5 k AIN7/ 400 mV 400 mV
J3
RP 1 0.1% IEXC IEXC V7 V8
AD8592 AD8592
to ADC A1
100 or AIN8 A2
1000 60 mA LCD DISPLAY 60 mA
Pt RTD 2 AIN6
Figure 11. LCD backlight drivers.
RX
AIN5/ V8 V9
Each half of the AD8592 op amp acts as a 60 mA current source to supply
3 IEXC to ADC the LCD backlight currents. The AD8592 can source and sink up to 250 mA,
3.3 V and the 100 nF capacitor ensures a soft startup.
1 k
RP 4 0.115 V AIN9 Hardware, Software, and User Interface
36
The complete circuit including software is available as the CN-0359 Circuits
RX =
V8 V9
1.5 k 2RP
from the Lab Reference Design. The circuit board, EVAL-CN0359-EB1Z,
V7 V8 comes preloaded with the code required to make the conductivity
Figure 9. Configuration for 2-wire RTD connection. measurements. The actual code can be found in the CN-0359 Design
Support Package, in the CN0359-SourceCode.zip file.
The 2-wire configuration is the lowest cost circuit and is suitable for less
critical applications, short RTD connections, and higher resistance RTDs The user interface is intuitive and easy to use. All user inputs are from a
such as Pt1000. dual function push button/rotary encoder knob. The encoder knob can be
turned clockwise or counterclockwise (no mechanical stop), and can also
Power Supply Circuits be used as a push button.
To simplify system requirements, all the required voltages (15 V and +3.3 V) Figure 12 is a photo of the EVAL-CN0359-EB1Z board that shows the LCD
are generated from a single 4 V to 7 V supply, as shown in Figure 10. display and the position of the encoder knob.
The ADP2300 buck regulator generates the 3.3 V supply for the board. The
design is based on the downloadable ADP230x Buck Regulator Design Tool.
Details regarding the selection and design of power supplies are available
at www.analog.com/ADIsimPower.
Use proper layout and grounding techniques to prevent the switching reg-
ulator noise from coupling into the analog circuits. See the Linear Circuit
Design Handbook, the Data Conversion Handbook, the MT-031 Tutorial,
and the MT-101 Tutorial for further details.
Figure 12. Photo of an EVAL-CN0359-EB1Z board showing the home
screen in measurement mode.
6V
PC
Data Sheets
Supply
AD8253 data sheet. ADA4638-1 data sheet.
AD8542 data sheet. ADG1211 data sheet.
J1
Conductivity J5
RS-485 to AD8592 data sheet. ADG1419 data sheet.
J2 USB
Cell
Adapter
ADA4000-1 data sheet. ADM3075 data sheet.
EVAL-CN0359-EB1Z
ADA4077-2 data sheet. ADP2300 data sheet.
RTD J3 J4 JTAG
ADA4528-2 data sheet. ADP1613 data sheet.
Figure 14. Test setup functional diagram. ADA4627-1 data sheet. ADuCM360 data sheet.
Robert Lee
Robert Lee [robert.lee@analog.com] has been an applications engineer
at Analog Devices since January 2013. Robert received his B.S.E.E. from Also by this Author:
University of Electronic Science and Technology of China (UESTC) in 2004 Complete Gas Sensor
and M.S.E.E. from UESTC in 2009. He has more than 10 years of embedded Circuit Using Nondis-
system design experience. persive Infrared (NDIR)
Volume 50, Number 4
Walt Kester
Walt Kester [walt.kester@analog.com] is a corporate staff applications
engineer at Analog Devices. During his many years at ADI, he has designed, Also by this Author:
developed, and given applications support for high speed ADCs, DACs, SHAs, Complete Gas Sensor
op amps, and analog multiplexers. An author of many papers and articles, he Circuit Using Nondis-
prepared and edited 11 major applications books for ADIs global technical persive Infrared (NDIR)
seminar series; topics include op amps, data conversion, power management, Volume 50, Number 4
sensor signal conditioning, mixed-signal circuits, and practical analog design
techniques. His latest book, The Data Conversion Handbook (Newnes), is a
nearly 1000-page comprehensive guide to data conversion. Walt has a B.S.E.E.
from NC State University and an M.S.E.E. from Duke University.
Share on
7
6
5 G = 11
4 G = 21
3 G = 31
2
0.25 5
Input
0.20 G = 31 V/V 4
G = 21 V/V
0.15 G = 11 V/V 3
0.10 2
0 0
0.05 1
0.10 2
0.15 3
0.20 4
0.25 5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Time (s)
Figure 3. Pulse response simulation results using the ADA4807 SPICE model. VS = 5 V,
R F = 10 k; G = 11 V/V, 21 V/V, and 31 V/V and R LOAD = 1 k.
Tina Collins [Tina.Collins@analog.com] is an applications engineer in the Linear Products and Technology Group
in Wilmington, MA. Her main interests are analog and mixed-signal design for high speed amplifiers. Tina joined
Analog Devices in 2001. Prior to becoming an applications engineer in 2013, she focused on developing and testing
high speed amplifiers. Tina received a B.S.E.E. from the University of Florida in 2001 and an M.S.E.E. from
Northeastern University in 2010.
Share on
Introduction
In high end factory automation applications such as gas and oil plants With a channel-to-channel isolation design, every channel needs dedicated
and power plants, the requirements of low EMI, small size, high reliability, power isolation and signal isolation. The isolation is one of the key limita-
and low cost are especially challenging for channel-to-channel isolated tions for input module channel density, EMI, cost, and reliability. In modern
designs. For this reason, the standard module implementation has a chan- designs, a digital isolator is used per channel for data isolation. A typical
nel density that is typically limited to only four or eight channels, with only digital isolator, such as theADuM141E, would have four isolated data
hundreds of volts of channel-to-channel isolation. channels in a 16-lead SOIC (6.2 mm 10mm) package. Power isolation
is still required per channel, though, so lets discuss three traditional
This article will briefly discuss isolation in the process control analog input approaches to power isolation: multitaps transformer, push-pull design,
module and the traditional approaches to achieving this. It then outlines and isolated dc-to-dc modules.
an alternative high density, easy to design, channel-to-channel isolation
analog input module architecture. Test results are included that show that Figure 2 shows a flyback isolation dc-to-dc architecture with a multitap
the 16-channel, 2.5 kV rms channel-to-channel isolation demo module transformer. A flyback converter drives the transformer to generate multiple
easily passes the EN55022, Class B isolation standard. outputs on the taps. Its a mature power architecture but has six major
disadvantages for process control applications, which are:
Isolation in the Process Control Analog Input Module
1. It needs a customized transformer with multitaps and a shield
Galvanic isolation is the principle of physically and electrically separat- to control EMI. This is hard to achieve in a small form factor with
ing two circuits, so that there is no direct conduction path but data and sufficient reliability.
power can still be exchanged. This is typically achieved using transform-
ers, optocouplers, or capacitors. Isolation is used to protect circuitry and 2. Only one channel could be used for a feedback control loop, meaning
human beings, break ground loops, and improve common-mode voltage that the other channels are more loosely regulated. This needs to be
and noise rejection performance. carefully evaluated to ensure reliable operation.
Typically process control inputs are either group isolated or channel-to- 3. Channel density is limited by specific transformer placement. For the
channel isolated (see Figure 1). For group isolation, a number of input power coming from each tap output, the transformer is placed as the
channels are grouped together to share a single isolation barrier, including center of the analog input module with each input channel laid out in
power isolation and signal isolation. This saves cost over channel-to-channel fanout sectors around the transformer, limiting the analog input module
isolation, but it limits the common-mode voltage difference between card channels to four or eight.
channels in the group, meaning they should all be placed in the same zone.
Channel-to-channel isolation, as in the right of Figure 1, is always favorable 4. Interference from one channel can be coupled into other channels
for its improved robustness. That said, it comes at a much higher cost per- through the coupling capacitors between the taps of the transformer.
channel, so this trade-off must be carefully assessed by plant builders.
5. The isolation voltage level. Multitap transformers can only achieve
Group Isolated Channel-to-Channel Isolated
hundreds of volts of channel-to-channel isolation unless special
AIN Ch. 1 AIN1 insulation materials or designs are incorporated, which significantly
A Ch. 2 increases the transformer cost.
. IN GND1 GND
. Group X
.
AIN Ch. M GNDx 6. The high cost of achieving UL/CSA certification for the customized
GND AIN2
transformer.
GND2 GND
AIN Ch. 1
A Ch. 2
. IN AINM
. Group Y
.
AIN Ch. M GNDy GND GNDm GND
I/O1A 3 22 I/O2A
Some of the drawbacks of this approach are the requirement for prereg-
I/O1B 4 21 I/O2B
ulation as well as additional components per channel. The transformer 4-Channel iCoupler Core
ADuM5410/ADuM5411/ADuM5412
selected must meet the isolation rating required. The preregulation, as well I/O1C 5 20 I/O2C
as a transformer, switches, and an LDO per channel take up board space I/O1D 6 19 I/O2D
and add cost. There is also a significant evaluation effort required to ensure VE1 7 18 VE2
that the regulation is sufficient under all conditions.
NIC 8 17 NIC
Switch Control
GND1 9 16 GNDISO
24 V 20% 7V 3.3 V PDIS 10 PCS 15 VSEL
Pre-
regulation LDO
VDDP 11 14 VISO
SPIData
P9
U3
E1 24 1 E5
VDD2 VDD1 3V3
E2 23 2 E6
3V3V9 GNDISO GND1
DIN9 E17 22 3 SPI2DIN
C4 1 F VOA VIA
SCLK9 E18 21 4 E25 SPI2SCLK C41 1 F C49 10 F
VOB VIB
C136 0.01 F CS_9 E19 20 5 E132 CS9
VOC VIC C219 0.010 F
DOUT_9 E20 19 6 E133 SPI2DOUT9
VID VOD
C50 1 F 18 7 E134
VE2 VE1
17 8 C209 1 F
C169 0.01 F NIC NIC C235 10 F
E9 16 9
GNDISO GND1 C220 0.010 F
15 10
C33 10 F VSEL PDIS
R202 R228 14 11 E30
VISO VDDP
16.9 k 10.2 k 13 12 E31
GNDISO GND1
E10
d
Figure 7. Floating stitching capacitor and overlapping stitching
capacitor.
For the floating stitching capacitor, two serial capacitors are built in, C1
and C2. The total capacitance is calculated by Equation 1.
C1 C2 lw1 lw2
C= , C1 = , C2 =
C1 + C2 d d
where:
is the permittivity of the PCB insulation material, 4.5 for FR4 material.
lw
C1 = ,
d
where:
is the permittivity of the PCB insulation material, 4 1011 F/m for FR4
material.
With the same material, area, and distance, the total capacitance value of
floating stitching is half that of the overlapping stitching, but the thickness
of the insulation material is doubled. Reinforced insulation, per IEC60950
2.10.6.4, requires 0.4 mm (15.74 mils) minimum insulation material thick-
ness in the interior layers, but basic insulation has no such requirement.
As the ADuM5411 only provides 2.5 kV rms basic isolation, an overlapping
stitching capacitor was chosen to maximize the capacitance. The thickness
of the interior layers has also been controlled to 5 mils for the same reason.
dB V/m
capacitors, KEMET C1812C102KHRACTU 3 kV, 150 pF, were mounted on 20
the board. Figure 12 shows the resultsit passed the EN55022 Class B
standard with 0.82 dB margin.
30
30 127 224 321 418 515 612 709 806 1000
Frequency in MHz
The results proved that stitching capacitors under the IC is a more effective
decoupling method than the safety capacitors.
Conclusion
Channel-to-channel isolation is often viewed as a design challenge in high
Figure 10. Stitching capacitors built into the PCB without safety capacitors. end process control systems. ADIsisoPower technology andiCoupler tech-
70
nology enables significant increases in channel density over traditional digital
Limit and power isolation approaches. They also greatly simplify the design task
Margin and can improve channel robustness and reliability. With stitching capacitor
built into the PCB, or safety capacitor mounted aside the PCB, EMI radiation
Class A
can be easily controlled to pass EN55022 Class B or Class A. Its a break-
Class B though in technology.
6
dB V/m
4 5
3
20
2 References
ADuM5411 data sheet. Analog Devices, Inc.
Van Yang
Van Yang [van.yang@analog.com] is a field application engineer at Analog Devices in Shanghai,
China. He joined ADI in 2015 to support regional medical and industrial customers in China.
Prior to joining ADI, Van worked at Texas Instruments as an FAE for four years. Van earned
his masters degree in communication and information systems from Huazhong University of
Science and Technology in Wuhan, in 2011. In his spare time he is a super fan of basketball and
enjoys hiking.
Songtao Mu
Songtao Mu [songtao.mu@analog.com] is a system applications engineer in Analog Devices
Industrial Automation Group. He is based in Shanghai, China. Prior to joining Analog Devices,
Songtao worked at Schneider Electric Co., Ltd., as a hardware design engineer for over eight
years. Songtao holds a bachelors degree in industrial automation from the Harbin institute of
Technology University, China.
Derrick Hartmann
Derrick Hartmann [derrick.hartmann@analog.com] is a system applications engineer in
the Industrial Automation Group at Analog Devices. He is based in Wilmington, MA. Prior to Also by this Author:
his current role, Derrick was a product applications engineer supporting Analog Devices
industrial DAC portfolio. Derrick holds a degree in electronic engineering from the University PLC Evaluation Board
of Limerick, Ireland. Simplifies Design of
Industrial Process
Control Systems
Share on
Common Signal Chain Design Pain Points Most ADC analog inputs, IN+ and IN, have no overvoltage protection
Figure 1 shows a typical signal chain used in building precision data acqui- circuitry apart from ESD protection diodes. In applications where the
sition systems. Applications that require precision data acquisition systems amplifier rails are greater than VREFand less than ground, it is possible
such as automated test equipment, machine automation, industrial, and for the output to go outside the input voltage range of the device. During an
medical instrumentation have common trends that are typically considered overvoltage event, the ESD protection diode between either analog input (IN+
technically conflicting. For example, system designers are forced to make or IN) pin to REF forward biases and shorts the input pin to REF, potentially
performance trade-offs to keep a tight system power budget or small area overloading the reference, causing damage to the device, or disturbing a
on the board to achieve high channel density. System designers of these reference that is shared among multiple ADCs. This results in having to add
precision data acquisition signal chains face common challenges in terms of protection circuitry like Schottky diodes to the ADC input to prevent overvolt-
driving the SAR ADC inputs, protecting ADC inputs from overvoltage events, age conditions from harming the ADC. Unfortunately, Schottky diodes could
reducing the system power with single supplies, and achieving higher add distortion and other errors due to leakage currents.
system throughput with low power microcontrollers and/or digital isolators.
Precision applications have different needs in terms of the processors that
interface to the ADC. Some applications need to be electrically isolated for
Signal Digital
Conditioning Ref
Ref
Isolator
safety reasons and use digital isolators between the ADC and processor to
Buf
Stage (Optional) achieve this. This choice of processor or need for isolation puts constraints
on the efficiency of the digital interface used to connect with the ADC.
FPGA/
Sensors
JFET/ Driver RC
ADC Micro-
Typically, lower end processors/FPGAs or lower power microcontrollers
In-Amp Stage Filter
controller have relatively low serial clock rates. This can result in lower than desired
throughput from the ADC because of a long ADC conversion time delay
Figure 1. Typical precision data acquisition signal chain. before the conversion result can be clocked out. Digital isolators can
also limit the maximum serial clock rate that can be achieved across the
Driving high resolution precision SAR ADCs has been traditionally a tricky isolation barrier due to propagation delays in the isolator limiting the ADC
issue because of the switched capacitor inputs. System designers need to throughput. In these scenarios an ADC that can achieve a higher through-
pay close attention to the ADC driver data sheet and look at the noise, dis- put rate without a significant increase in serial clock rate is desirable.
tortion, input/output voltage headroom/footroom, bandwidth, and settling
time specifications. Typically, high speed ADC drivers are required that are
wide bandwidth, low noise, and high power in order to settle the switched
AD4000/AD4003 Precision SAR ADC Family Solves
capacitor kickback of the SAR ADC inputs within the available acquisition Common Design Challenges
time. This significantly reduces the options available for amplifiers to The AD4000/AD4003 family is a fast, low power, single-supply, 16-/18-bit
drive the ADC and results in significant performance/power/area trade- precision ADC based on a SAR architecture.
Low Power Figure 3 shows the input current of the AD4000/AD4003 ADC with high-Z
16 mw @ 2 MSPS Small Footprint mode enabled/disabled. The low input current makes the ADC a lot easier
80 W @ 10 kSPS 3 mm 3 mm
to drive than traditional SAR ADCs available in the market even with high-Z
mode disabled. If you compare the input current in Figure 3 with high-Z
mode disabled against that of the previous generationAD7982ADC, the
AD4003 has reduced the input current by 4 at 1MSPS. The input current
reduces further to submicroampere range when high-Z mode is enabled.
High-Z mode should be disabled for input frequencies above 100kHz or
when multiplexing the input.
12.0
9.0
0.0
3.0
High Throughput
2 MSPS 6.0
25C High-Z Enabled
Figure 2. AD4000/AD4003 ADC key benefits. 9.0
25C High-Z Disabled
12.0
AD4000/AD4003 ADC Ease of Use Features
15.0
5 4 3 2 1 0 1 2 3 4 5
Long Acquisition Phase
Input Differential Voltage (V)
The AD4000/AD4003 ADC features a very fast conversion time of 290 ns
and the ADC returns back to the acquisition phase 100 ns before the end Figure 3. AD4003 ADC input current vs. input differential voltage
with high-Z enabled/disabled.
of the ongoing conversion process. SAR ADC cycle time is comprised of
conversion and acquisition phases. During the conversion phase, the ADC As shown in Figure 4, the AD4000/AD4003 ADC allows a choice of lower
capacitor DAC is disconnected from the ADC inputs to perform the SAR power/bandwidth precision amplifiers with a lower RC filter cutoff to drive
conversion. The inputs are reconnected during the acquisition phase, and the ADC, removing the need for dedicated high speed ADC drivers, saving
the ADC driver must settle the inputs to the correct voltage before the next system power, size, and cost in precision, low bandwidth applications
conversion phase begins. A longer acquisition phase reduces the settling (signal bandwidths <10 kHz). Ultimately, the AD4000/AD4003 allows the
requirement on the driving amplifier and allows a lower RC filter frequency amplifier and RC filter in front of the ADC to be chosen based on the signal
cutoff, which means a higher noise and/or lower power/bandwidth ampli- bandwidth of interest and not based on the settling requirements of the
fier can be tolerated. A larger value of R can be used in the RC filter with switched capacitor SAR ADC inputs.
a corresponding smaller value of C, reducing amplifier stability concerns
without impacting distortion performance significantly. A larger value of R Signal Digital
Ref
helps to protect the ADC inputs from overvoltage conditions. It also results Conditioning Ref
Buf
Isolator
in reduced dynamic power dissipation in the amplifier. Stage (Optional)
FPGA/
JFET/ Driver RC
Sensors ADC Micro-
In-Amp Stage Filter
controller
THD (dB)
typical THD with high-Z enabled for a 2.27MHz RC bandwidth and 1kHz
100
input signal. THD is approximately 10dB better with high-Z mode enabled
even for large R values greater than 200. SNR holds up close to 104
99dB even with a very low RC filter cutoff. 108 ADA40771, 2 MSPS, High-Z Disabled
ADA40771, 2 MSPS, High-Z Enabled
112 ADA46101, 2 MSPS, High-Z Disabled
With high-Z enabled, the ADC will consume around 2 mW/MSPS extra ADA46101, 2 MSPS, High-Z Enabled
power, but this would be still significantly lower than using dedicated ADC 116 ADA40842, 2 MSPS, High-Z Disabled
drivers like theADA4807-1and results in PCB area and bill of material ADA40842, 2 MSPS, High-Z Enabled
120
savings. For most systems, the front end usually limits the overall ac/dc 470 pF 470 pF 180 pf 180 pf 180 pf
performance achievable by the signal chain. Its evident from the selected 1.3 k 680 680 390 200
260.482 kHz 497.981 kHz 1.3 MHz 2.27 MHz 4.42 MHz
precision amplifiers data sheet in Figure 5 and Figure 6 that its own noise Capacitor (pF), Resistor (), and RC Filter Bandwidths (Hz)
and distortion performance dominates the SNR and THD specification at
a certain input frequency. However, the AD4003 ADC with high-Z mode
Figure 6. THD vs. RC bandwidths using ADA4077, ADA4084, and
allows a greatly expanded choice of driver amplifier including precision
ADA4610 precision amplifiers.
amplifiers used in signal conditioning stages along with more flexibility
in the RC filter choice. For example, when the AD4003 ADCs high-Z
is enabled and using theADA4084-2driver amplifier with a wideband Figure 7(a) shows that system designers can use the 2.5 lower pow-
input filter of 4.42 MHz, the SNR performance is about 95 dB. With more er ADC driver ADA4077 (vs. the ADA4807), and the AD4003 ADC still
aggressive filtering of the ADC drivers noise using a 498 kHz filter, SNR achieves SINAD of about 97dB (3dB better than the AD7982 ADC) when
improves by 3 dB, to 98 dB. The AD7982 ADCs SNR performance at lower high-Z mode is disabled. Even with a wider RC bandwidth of 2.9 MHz, the
RC cutoff is degraded because the ADC input does not settle the kickback ADA4077 amplifier cannot drive the AD7982 ADC directly and achieve op-
within its short acquisition time. timum performance. The driver cannot settle the ADC kickback within the
available acquisition time with aggressive filtering at a lower RC bandwidth
100
98
cutoff and hence the ADC SINAD performance is degraded. The AD4003
96
ADCs switched capacitor kickback with either high-Z mode disabled or
94 enabled is much reduced and the acquisition time is 2.5 longer at 1 MSPS
92 and hence its SINAD performance is still significantly better than that of
90 AD7982 ADC.
88
With high-Z mode enabled, the AD4003 ADCs SINAD performance is better
SNR (dB)
86
84 using both ADC drivers at lower RC filter cutoff, which helps remove more
82 wideband noise coming from the upstream signal chain components when
80 ADA4077-1, 2 MSPS, High-Z Disabled the signal bandwidth of interest is low. Without high-Z mode enabled there
78 ADA4077-1, 2 MSPS, High-Z Enabled is a trade-off between RC filter cutoff and SINAD performance.
76 ADA4610-1, 2 MSPS, High-Z Disabled
ADA4610-1, 2 MSPS, High-Z Enabled
74
ADA4084-2, 2 MSPS, High-Z Disabled Span Compression
72 ADA4084-2, 2 MSPS, High-Z Enabled
70 The AD4000/AD4003 ADC includes a span compression mode, which is
470 pF 470 pF 180 pf 180 pf 180 pf useful for systems that only have a single positive supply to power the
1.3 k 680 680 390 200
260 kHz 498 kHz 1.3 MHz 2.27 MHz 4.42 MHz SAR ADC drivers. It eliminates the ADC drivers need for a negative supply
Capacitor (pF), Resistor (), and RC Filter Bandwidths (Hz) while preserving the full resolution of the ADC, saving power and reducing
Figure 5. SNR vs. RC bandwidths using ADA4077, ADA4084, and power supply design complexity. As shown in Figure 8, the ADC performs
ADA4610 precision amplifiers. a digital scaling function that maps zero-scale code from 0 V to 0.1 V
2.5 Lower Power
100 100
98 98
96 96
94 94
92 92
90 90
88 88
86 86
84 84
82 82
80 80
SINAD (dB)
SINAD (dB)
78 78
76 76 Higher Performance
74 74 Maintained at Lower
72 72 RC Bandwidth
70 70
68 68
66 66
64 64
62 62
60 AD7982/ADA4077: 1 MSPS 60 AD7982/ADA4077: 1 MSPS
58 AD4003/ADA4077: 1 MSPS 58 AD4003/ADA4077: 1 MSPS
56 AD7982/ADA4077: 1 MSPS 56
54 54 AD7982/ADA4077: 1 MSPS
52 AD4003/ADA4077: 1 MSPS 52 AD4003/ADA4077: 1 MSPS
50 50
470 pF 470 pF 180 pf 180 pf 180 pf 470 pF 470 pF 180 pf 180 pf 180 pf
1.3 k 680 680 390 200 1.3k 680 680 390 200
260.482 kHz 497.981 kHz 1.3 MHz 2.27 MHz 4.42 MHz 260.482 kHz 497.981 kHz 1.3 MHz 2.27 MHz 4.42 MHz
Capacitor (pF), Resistor (), and RC Filter Bandwidths (Hz) Capacitor (pF), Resistor (), and RC Filter Bandwidths (Hz)
(a) High-Z Mode Disabled (b) High-Z Mode Enabled
Figure 7. AD4003 ADC and AD7982 ADC amplifier driver comparison using ADA4077 and ADA4807: SINAD vs. RC bandwidths for high-Z mode
disabled and enabled (F S= 1 MSPS, f IN= 1 kHz).
Digital Output
AD4000/AD4003 ADC Performance
5V
Analog
+FSR Operating from a 1.8 V supply, the AD4000/AD4003 ADC consumes
Input typically 14mW/16mW at 2MSPS and offers superior linearity of
0.41 V
All
ADC
2N Codes 1.0 LSB (3.8 ppm) max and guaranteed 18 bits no missing codes.
Figure 11 shows the AD4003 ADCs typical INL vs. code performance. The
FSR
AD4003 ADC achieves better SINAD performance than the AD7982 ADC
Figure 8. AD4000/AD4003 ADC span compression operation.
over a wide range of input frequencies up to Nyquist (Figure 12), enabling
system designers to develop wider bandwidth, higher precision instrumen-
Overvoltage Clamp tation equipment. The AD4000/AD4003 ADC is available in a small 10-lead
In applications where the amplifier rails are greater than VREFand less than footprint (3 mm 3 mm, LFCSP and 3 mm 5 mm, MSOP options), and is
ground it is possible for the output to violate the input voltage range of the pin-compatible with the AD798x/AD769x ADC family.
part. When positive input is over-ranged, current flows through D1 into
1.0
REF (see Figure 9), disturbing the reference. Even worse it could pull the
reference above the absolute maximum reference value and hence could 0.8 +125C
+25C
damage the part. 0.6 40C
0.4
When the analog input exceeds the reference voltage by ~400 mV, the
AD4000/AD4003 ADCs internal clamp circuit will turn on and the current 0.2
INL (LSB)
will flow through the clamp into ground, preventing the input from rising 0
further and potentially causing damage to the device.
0.2
REF
0.4
0.6
D1
0 V to 15 V REXT IN+/IN RIN CIN
0.8
1.0
0 32768 65536 98304 131072 163840 196608 229376 262144
CEXT CPIN CODE
VIN D2 Clamp
Figure 11. AD4003 ADC INL vs. code.
101
GND
AD4003
98
Figure 9. AD4003 ADC equivalent analog input circuit.
95
As shown in Figure 9, the AD4000/AD4003 ADCs internal overvoltage
clamp circuit with a larger external resistor (REXT=200 ) eliminates the
SINAD (dB)
92
need for external protection diodes (and hence the need for additional
board space). The clamp turns on before D1 and can sink up to 50 mA of 89
current. The clamp prevents damage to the device by clamping the input
voltage to a safe operating range and avoids disturbance of the reference, 86
which is particularly important for systems that share the reference among
multiple ADCs. 83 AD4003
AD7982
80
Efficient Digital Interface 1 10 100 1k
The AD4000/AD4003 ADC has a flexible digital serial interface that offers Frequency (kHz)
seven different modes with register programmability. Its turbo mode allows Figure 12. AD4003 ADC vs. AD7982 ADC SINAD vs. input frequency.
the user to start clocking out the previous conversion results while the ADC
is still converting, as shown in Figure 10. A combination of short conversion The AD4000/AD4003 ADC powers down automatically at the end of each
time and turbo mode allows a lower SPI clock rate and simplifies the isola- conversion phase; therefore, its power scales linearly with the throughput
tion solution, resulting in reduced latency requirements on digital isolators as shown in Figure 13. This feature makes the device ideal for low sam-
and broadening a choice of processors including lower end processors/FP- pling rates (even down to a few Hertz) and battery-powered portable and
GAs or low power microcontrollers with relatively low serial clock rates. For wearable systems. Even in low duty cycle applications, the first conversion
example, the AD4003 ADC can use a 2.5 slower SPI clock rate (25MHz result is always valid.
vs. 66MHz) than the AD7982 ADC when running at 1 MSPS. The user can
write/read back the register bits to enable the AD4000/AD4003 ADCs ease
tCYC
CNV
TACQ
tCONV L tSCK
tSCK
tQUIET2
tQUIET1
SCK 1 2 3 16 17 18
tHSDO tSCKH
Automated Test
10
0
0 200 400 600 800 1k 1.2k 1.4k 1.6k 1.8k 2k
Throughput (kSPS)
System Applications
A combination of ease of use features and high performance, small
footprint, and low power makes the AD4000/AD4003 ADC family a great
solution for many precision control and measurement system applications
Optical Avionics
as shown in Figure 14. The AD4000/AD4003 ADC reduces measurement Communications Measurement
uncertainty, increases repeatability, allows high channel density, and in- and Control
creases the throughput efficiency of automated test equipment, automated
machine control equipment, and medical imaging equipment. This ADC is a
good fit for systems that demand higher frequency performance to capture Medical Imaging CT
fast transients and time of flight information, such as power analyzer and and Digital X-Ray
mass spectrometer applications. Figure 14. AD4000/AD4003 ADC end system applications.
Share on
Question: It is far easier to obtain matched NPN BJTs than matched PNPs, although
How do I design a circuit with an accurate current output into a nonlinear the latter are available. A current mirror of a current from positive supply
load resistance? to a grounded source with the mirrored current to a grounded load may be
made with matched PNP transistorsbut there are also a couple of fast
current mirrors with a dynamic range of 106 (1,000,000:1), the ADL5315,
andADL5317.2
For a voltage input you simply need an op amp, a transistor (FET or BJT),
and a resistor. The input is applied to the op amp noninverting input, the op
amp drives the transistor gate/base, the resistor is connected to ground,
and to the op amp inverting input and the source/emitter, and the output
current flows in the drain/collector.
If the input to the amplifier is also a stiff current source (that is, a current The short article mentioned below describes the circuits mentioned above
whose value does not vary with loading), a current mirror using two in more detail, and with diagrams, but the basic principles of accurate
matched bipolar junction transistors (BJTs), on a single chip to ensure current output amplifiers are simple.1
temperature matching, is often all that is needed. The current is applied to
When I started to write this article I was going to describe how a precision
both bases and one collector, the two emitters are grounded and an equal
two terminal floating current source could be built with an op amp, a
current flows in the other collectorslightly more complex arrangements
precision voltage reference, three resistors, and a capacitor, but since
may improve performance but the basic circuit is often adequate.
Analog Devices and Linear Technology have announced that they are
James Bryant
James Bryant [ james@jbryant.eu] was a European applications manager
at Analog Devices from 1982 to his retirement in 2009 and still writes and Also by this Author:
consults for the company. He holds a degree in physics and philosophy Who Killed the
from the University of Leeds and is also C.Eng., EurEng., MIET, and an FBIS. In Component?
addition to his passion for engineering, James is a radio ham and holds the Volume 50, Number 4
call sign G4CLF.
Conclusion
The AD4000/AD4003 ADC family enables designers to solve the system-level
technical challenges for their high precision data acquisition system
without making significant trade-offs, reducing the total system design
time. The AD4000/AD4003 ADCs high performance increases measure-
ment accuracy and its small footprint coupled with low system-level
thermal dissipation enables higher density.
Maithil Pachchigar
Maithil Pachchigar [maithil.pachchigar@analog.com] is an applications
Also by this Author:
engineer in the Instrumentation, Aerospace and Defense business unit at
Analog Devices in Wilmington, MA. Since joining ADI in 2010, he has been Integrated Multi-
focused on the precision ADC product portfolio and customers in the plexed Input ADC
Solution Alleviates
instrumentation, industrial, healthcare, and energy segments. Having worked Power Dissipation and
in the semi-conductor industry since 2005, he has published numerous Increased Channel
technical papers. Maithil received his B.E. in electronics engineering degree Density Challenges
from S.V. National Institute of Technology, India, in 2003, M.S.E.E. degree from Volume 50, Number 1
San Jose State University in 2006, and M.B.A. degree from Silicon Valley
University in 2010.
Alan Walsh
Like Analog Dialogue to discover the latest articles and author insights.
facebook.com/analogdialogue
Analog Devices, Inc. Analog Devices, Inc. Analog Devices, Inc. Analog Devices, Inc. 2017 Analog Devices, Inc. All rights reserved. Trademarks and
Worldwide Headquarters Europe Headquarters Japan Headquarters Asia Pacific Headquarters registered trademarks are the property of their respective owners.
Ahead of Whats Possible is a trademark of Analog Devices.
Analog Devices, Inc. Analog Devices GmbH Analog Devices, KK Analog Devices M02000504-0-1/17
One Technology Way Otl-Aicher-Str. 60-64 New Pier Takeshiba 5F, Sandhill Plaza
P.O. Box 9106 80807 Mnchen South Tower Building 2290 Zuchongzhi Road analog.com
Norwood, MA 02062-9106 Germany 1-16-1 Kaigan, Minato-ku, Zhangjiang Hi-Tech Park
U.S.A. Tel: 49.89.76903.0 Tokyo, 105-6891 Pudong New District
Tel: 781.329.4700 Fax: 49.89.76903.157 Japan Shanghai, China 201203
(800.262.5643, U.S.A. only) Tel: 813.5402.8200 Tel: 86.21.2320.8000
Fax: 781.461.3113 Fax: 813.5402.1064 Fax: 86.21.2320.8222